Near ultraviolet photolysis of ammonia and
methylamine studied by H Rydberg atom
photofragment translational spectroscopy
B y M. N. R. A s h f o l d, R. N. D i x o n, M. K o n o, D. H. M o r d a u n t†
a n d C. L. R e e d
School of Chemistry, University of Bristol, Bristol BS8 1TS, UK
H(D) Rydberg atom photofragment translational spectroscopy has been used to provide new insights into the primary photochemistry of methylamine, ammonia and
various of their respective isotopomers following excitation at wavelengths in the
near ultraviolet (UV). The bimodal appearance of the total kinetic energy release
(TKER) spectra associated with H atom production in the near UV photolysis of
methylamine is consistent with there being both ‘dynamical’ (high TKER) and ‘statistical’ (slower) contributions to the total H + CH3 NH dissociation yield. Both contributions arise as a result of one H atom tunnelling through (or passing over) an
earlier barrier in the N–H dissociation coordinate of the à state potential energy
surface and then evolving into the region of the conical intersection connecting the
à state and ground-state surfaces. ‘Dynamical’ energy disposal is associated with
those molecules which pass directly through this conical intersection en route to the
ground-state (H + CH3 NH(X̃)) asymptote, whilst the ‘statistical’ contribution is attributed to those molecules that ‘miss’ the conical intersection on the first traversal
and only make the à → X̃ transfer on a later encounter. This interpretation has
inspired further consideration of the form of the TKER spectra derived from TOF
measurements of the H/D atom products arising in the dissociation of various isotopomers of ammonia following excitation to the 00 and 21 levels of their respective
à states. A similar model which associates ‘dynamical’ energy disposal with those
molecules that pass through the Ã/X̃ conical intersection during bond extension,
and ‘statistical’ kinetic energy release with those that transfer during N–H(D) bond
compression, appears to provide a qualitative explanation for the way the observed
H and/or D atom yields and their associated TKER spectra vary with excitation
wavelength (00 versus 210 band excitation) and isotopic composition.
1. Introduction
Photodissociation plays a key role in cycling gas phase molecules in the atmospheres
of Earth and of the various planets. Recent laboratory studies of the photochemistry
of, for example, ozone (Miller et al. 1994; Ball et al. 1995; Takahashi et al. 1996)
and methane (Mordaunt et al. 1993) are providing new insights into the chemistry
prevailing within our own stratosphere (Mack et al. 1996) and in the atmospheres of
† Present address: Department of Chemistry, University of California Berkeley, CA 94720, USA.
Phil. Trans. R. Soc. Lond. A (1997) 355, 1659–1676
1659
Printed in Great Britain
c 1997 The Royal Society
TEX Paper
1660
M. N. R. Ashfold and others
various of the moons of Jupiter and Saturn, e.g. Titan (Lara et al. 1994), respectively.
Here we present the results of experimental studies of the primary photochemistry of
ammonia (NH3 ) and methylamine (CH3 NH2 ) following photoexcitation to their respective first excited singlet electronic states. The observable in these experiments is
a high resolution time-of-flight (TOF) spectrum of the atomic hydrogen photoproducts; the technique relies on the remarkable stability of the high-n Rydberg states
of atomic H and D.
The first (Ã1 A002 − X̃ 1 A01 ) absorption band of ammonia spans the wavelength range
170–220 nm and is associated with promotion of one of the nitrogen lone pair electrons to an excited orbital with mixed Rydberg (3s)–antibonding valence (σ ∗ ) character (Douglas 1963; Runau et al. 1977; Vaida et al. 1984; McCarthy et al. 1987; Rosmus
et al. 1987; Ashfold et al. 1996). The absorption spectrum exhibits a clear progression
in ν20 , the excited state out-of-plane (umbrella) bending mode, reflecting the planar ←
pyramidal change in minimum energy configuration upon electronic excitation. The
corresponding ÖX̃ absorption system in methylamine (190 nm < λ < 240 nm) also
exhibits regular but diffuse vibronic structure which has been assigned (Tsuboi et al.
1969) to progressions involving the NH2 wagging (ν9 ) and CH3 rocking (ν14 ) modes.
However, a more recent analysis (Taylor et al. 1995) of resonance enhanced multiphoton ionization (REMPI) spectra of jet-cooled samples of the four isotopomers
CH3 NH2 , CH3 ND2 , CD3 NH2 and CD3 ND2 suggests that the second Franck–Condon
active mode is actually the ν4 (NH2 scissors) vibration. Such an interpretation is
appealing in that it implies that electronic excitation induces a trend towards sp2
hybridization at the N centre analogous to that in ammonia.
The dissociation dynamics of NH3 molecules following photoexcitation to their
Ã1 A002 excited state have been much studied, with the result that this system is often
cited as a textbook example of a molecular photofragmentation process (Ashfold et
al. 1996). Linewidth studies show that the excited state lifetime varies sensitively
with H/D isotopic substitution and with the v2 quantum number; in all isotopomers,
the level with v2 = 1 is the most long lived. Such behaviour has been rationalized
in terms of vibrational predissociation on the à state potential energy surface. Ammonia molecules in their à state are quasi-bound; dissociation from the two lowest
vibrational levels (i.e. those with v2 6 1) occurs predominantly, if not exclusively,
via H atom tunnelling through a barrier in the exit channel leading to the products
H + NH2 . This barrier arises as a natural consequence of the Rydberg (3s) → antibonding valence (σ ∗ ) orbital evolution as one N–H bond is extended (Runau et al.
1977; Rosmus et al. 1987); the barrier height is least (and tunnelling consequently
most facile) at planar geometries (Ashfold et al. 1985; McCarthy et al. 1987; Rosmus
et al. 1987). Ã state ammonia molecules with v20 > 1, or with excitation in vibrational
modes other than v2 , have sufficient internal energy that, if redistributed appropriately, dissociation can occur by passage over (rather than through) the exit channel
barrier on the à state surface.
Planarity, or otherwise, also influences the asymptotic product correlations. For
planar configurations, the X̃ state of ammonia correlates with the excited products
H + NH2 (Ã2 A1 ), whilst the parent à state correlates with the ground-state products
H + NH2 (X̃ 2 B1 ). Away from planarity, the X̃ and à states of ammonia have the same
(A0 ) electronic symmetry. Thus these two potential energy surfaces can only ‘cross’
at planar geometries and, consequently, there is a conical intersection (CI) between
the à and X̃ state surfaces, at planar geometries, in the H–NH2 exit channel. As a
result, the parent à state surface shows a deep well in this exit channel, which acts
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
1661
as a funnel, accelerating many of the dissociating trajectories through this narrow
region of configuration space onto the X̃ state surface and thence to the ground-state
products. Energetic considerations dictate that ammonia photolysis at λ > 206 nm
can only lead to ground-state products; the excited H + NH2 (Ã) products become
increasingly important at shorter photolysis wavelengths, accounting for ca. 30% of
the total dissociation yield at 193 nm (Biesner et al. 1989).
Previous H atom photofragment translational spectroscopy (PTS) experiments involving the ammonia molecule have highlighted the profound influence of this CI on
the quantum state population distributions within the resulting NH2 (X̃) fragments.
A substantial fraction of these products are formed with little vibrational excitation,
but with considerable rotational excitation, specifically distributed in the form of
a-axis rotation (Biesner et al. 1988, 1989; Ashfold et al. 1990; Dixon 1996; Mordaunt
et al. 1996a, b). This very non-statistical energy disposal is a natural consequence of
the CI amplifying any out-of-plane bending motion in the parent NH3 molecule into
a-axis rotation of the fragment. The later studies also identify an unresolved, ‘statistical’ contribution to the total H + NH2 (X̃) product yield, the relative importance
of which depends sensitively upon H/D isotopic substitution and the parent vibrational level. This has been attributed to, and modelled in terms of, Ã → X̃ internal
conversion (IC), followed by unimolecular decay (Mordaunt et al. 1996b), but the
mechanism of the IC process remains an open question.
Methylamine photodissociation has received much less attention but inevitably,
given the greater number of degrees of freedom, it has a potentially richer photochemistry. Thermochemical considerations suggest the possibility of seven different
fragmentation channels following absorption of a 220 nm photon (the wavelength corresponding to the peak of the first absorption band) (Hwang et al. 1990; Reed et al.
1996). Classical photochemical experiments (Michael et al. 1963), together with more
recent PTS studies (Waschewsky et al. 1995; Reed et al. 1996), suggest that channels (1.1)–(1.4) all contribute to the primary UV photochemistry of methylamine,
with channel (1.4) contributing more than 75% of the total dissociation yield:
CH3 NH2 + hn → CH2 = NH + H2 ,
→ CH3 + NH2 ,
→ CH2 NH2 + H,
→ CH3 NH + H,
∆H0 = 10600 cm−1 [21],
D0 = 29300 cm
(1.1)
−1
[21],
(1.2)
−1
[21],
(1.3)
−1
[22].
(1.4)
D0 = 30600 cm
D0 = 35900 cm
Interpreting this finding, and the deduced form of the energy partitioning in the
various dissociation products, relies heavily on the theoretical work of Kassab et al.
(1983) and Dunn & Morokuma (1996). The latter workers present cuts through ab
initio potential energy surfaces for the à and X̃ states of methylamine. These show
that the à state potential for both C–N and N–H bond fission (channels (1.2) and
(1.4), respectively) is, in each case, characterized by a small barrier at short range
and, at larger bond extensions, by a conical intersection with the X̃ state surface
at geometries for which the NH2 moiety and one C–H bond are coplanar. Thus the
topology of the à and X̃ state surfaces of methylamine, along the N–H dissociation coordinate, is qualitatively similar to that in ammonia. Dunn & Morokuma
(1996) also calculate that: (i) the barrier to C–N bond fission on the à state surface
is somewhat larger than that for N–H bond rupture (consistent with the observed
dominance of channel (1.4) following near UV excitation of CH3 NH2 ); and (ii) C–
H bond extension in the à state correlates adiabatically with the excited products
Phil. Trans. R. Soc. Lond. A (1997)
1662
M. N. R. Ashfold and others
H + CH2 NH2 (Ã)—a process which is endoergic for all λ & 190 nm. Thus any experimentally observed products attributable to dissociation channel (1.3) must arise
as a result of IC from the initially populated à state and subsequent dissociation
on the ground-state surface. Such is broadly consistent with experimental findings
(Waschewsky et al. 1995): the total kinetic energy release (TKER) spectra associated
with the H + CH3 NH and CH3 + NH2 product channels both peak away from zero
kinetic energy, consistent with dissociation on a surface for which there is a barrier
to the reverse reaction, whilst the TKER spectrum attributed to the H + CH2 NH2
channel is deduced to peak near zero. Recent work from our own group (Reed et al.
1996), however, indicates that these ‘slow’ H atoms actually also arise via dissociation
channel (1.4).
Here we consider high-resolution TKER spectra of the H(D) atom photofragments
arising in the photodissociation of various of the H/D isotopomers of ammonia and
methylamine following monochromatic excitation in the near UV. Much of the ammonia data has been presented previously (Mordaunt et al. 1996a, b); the observations and much of the interpretation that derives therefrom will only be summarized
briefly. The TKER spectra of the H(D) atoms resulting from methylamine photolysis appear bimodal, suggesting contributions both from direct bond fission and from
a slower, more ‘statistical’ unimolecular decay process. The deduced importance of
dissociation channel (1.4) as the source of the ‘slow’ H(D) atoms, arising via unimolecular decay after IC from the initially populated à state of methylamine, is
most readily accommodated by assuming that IC occurs in the region of the Ã/X̃
conical intersection in the N–H exit channel. Given this conclusion, we consider the
evidence for and against such an explanation for the ‘statistical’ component of the
H(D) atom TKER spectrum arising in the UV photolysis of ammonia.
2. Experimental
The apparatus used to measure H-atom photofragment translational spectra and
the ‘tagging’ of the H atoms has been described previously (Morley et al. 1992, 1993;
Mordaunt et al. 1996a, b; Reed et al. 1996). A pulsed skimmed supersonic molecular
beam of the chosen parent molecule (seeded in Ar) is crossed, orthogonally, by the
output of three dye lasers. Laser 1, the output of which is frequency doubled and
then mixed to produce light in the range 200 < λ < 240 nm, induces photolysis. The
resulting H(D) atom photofragments are then ‘tagged’ via a two-colour two-photon
(121.6 and 365.7 nm) excitation to a high-n Rydberg state. The detected H(D) atoms
fly a known field-free distance along the third orthogonal axis and are field-ionized
immediately before detection. The detector output is amplified and sent to a digital
storage oscilloscope for display. This transient TOF spectrum is transferred to a
computer via a GPIB interface and allowed to accumulate for (typically) 104 laser
shots. Energy and momentum conservation allows conversion of the accumulated
TOF spectra to TKER spectra as described in the publications cited above.
3. Results and discussion
(a ) Ammonia photolysis: the ‘dynamical’ channel
The observable in all of these experiments is the H atom yield arriving at a detector
located a known distance, d, from the photolysis volume, recorded as a function of
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
excitation via
000
1663
bands
(a)
(d)
(b)
(e)
TKER / cm–1
(c)
TKER / cm–1
Figure 1. TKER spectra (all normalized to the same peak intensity and all plotted on a common
energy scale) for the fragments resulting from photolysis of jet-cooled samples of the various
isotopomers of ammonia: (a) NH3 → H + NH2 excited at 46 200 cm−1 ; (b) NHD2 → H + ND2
at 46 529 cm−1 ; (c) ND3 → D + ND2 at 46 708 cm−1 ; (d) NH2 D → H + NHD at 46 359 cm−1 ;
and (e) NHD2 → D + NHD at 46 529 cm−1 , i.e. at the peaks of their respective ÖX̃ origin
absorptions, with, in each case, ε aligned perpendicular to the TOF axis.
time after the photolysis laser pulse. Signal arriving with a particular TOF, tH , can
be related to the corresponding photofragment TKER, Ekin,total , via the relationship
2
mH
d
1
,
(3.1)
Ekin,total = Ekin,H + Ekin,R = 2 mH 1 +
mR
tH
where R is the fragment partnering the detected H atom (e.g. NH2 in the case of NH3
photolysis, or a fragment having chemical formula CH4 N in the case of methylamine
photodissociation). Clearly, an equivalent expression holds when detecting D atom
fragments arriving with a TOF tD .
Figures 1 and 2 show representative TKER spectra obtained by monitoring H
and/or D atoms resulting from photolysis of jet cooled samples of the various isotopomers of ammonia following excitation within their respective ÖX̃000 and 210
Phil. Trans. R. Soc. Lond. A (1997)
1664
M. N. R. Ashfold and others
excitation via 210 bands
(a)
(d)
(b)
(e)
TKER / cm–1
(c)
TKER / cm–1
Figure 2. TKER spectra (all normalized to the same peak intensity and all plotted on a common
energy scale) for the fragments resulting from photolysis of jet-cooled samples of the various
isotopomers of ammonia: (a) NH3 → H + NH2 excited at 47 069 cm−1 ; (b) NHD2 → H + ND2 at
47 264 cm−1 ; (c) ND3 → D + ND2 at 47 362 cm−1 ; (d) NH2 D → H + NHD at 47 155 cm−1 ; and
(e) NHD2 → D + NHD at 47 264 cm−1 , i.e. at the peaks of their respective ÖX̃ 210 absorptions,
with, in each case, ε aligned perpendicular to the TOF axis.
bands. The sharp structure evident in these spectra is assignable in terms of formation of NH2 /NHD/ND2 fragments carrying little vibrational excitation, but with
a wide spread of rotational energies; in all cases, however, this rotational energy is
concentrated in the form of a-axis rotation of the fragment (Biesner et al. 1988, 1989;
Ashfold et al. 1990; Dixon 1996; Mordaunt et al. 1996a, b). Similar sharp structure
was observed in previous (lower resolution) H atom PTS studies of NH3 photolysis at
still shorter wavelengths, λ > 193 nm, i.e. for photoexcitation of à state vibronic levels with v20 6 6 (Biesner et al. 1988). These spectra derive from TOF measurements
made with the ε vector of the photolysis laser radiation aligned perpendicular to the
TOF axis. Each photolysis has been investigated as a function of θ, the angle between
ε and the H(D) atom recoil velocity vector, v. Such studies (Mordaunt et al. 1996a)
reveal that, although the angular dependence of the total H(D) atom flux is rather
isotropic, ε and v are nonetheless strongly correlated, but in a manner that varies
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
1665
Figure 3. Expanded view of the TKER spectrum of the D + NHD fragments resulting from photodissociation of NHD2 at 47264 cm−1 . The solid curve shows the TKER spectrum expected
for a wholly ‘statistical’ energy partitioning such as might occur for dissociation of highly vibrationally excited NHD2 (X̃) molecules formed by IC from the initially photoexcited à state.
with the particular v, N rovibrational state in which the partner NH2 /NHD/ND2
fragment is formed. H(D) atoms formed in association with rotationally ‘cold’ fragments show with greatest probability when θ = 90◦ (i.e. as in figures 1 and 2), whilst
those fragments carrying the highest levels of a-axis rotational excitation show most
clearly when ε is parallel to the TOF axis (i.e. when θ = 0◦ ).
The following explanation for such observations is phrased specifically in terms of
NH3 photodissociation, but analogous arguments hold for the lower symmetry mixed
isotopomers also. Ã state ammonia molecules have a planar equilibrium geometry
and the ÖX̃ transition moment, µparent , lies perpendicular to this plane. Following
excitation to the 00 and 21 levels, the dissociating molecules tunnel through the
exit channel barrier and emerge with a range of out-of-plane configurations (centred
about planarity) and momenta. Those that approach the CI with (essentially) planar
geometries pass straight through to the ground-state surface. There are no torques
acting to generate out-of-plane motion and the H and NH2 (X̃) fragments separate in
the plane perpendicular to µparent and thus to ε, i.e. the NH2 (X̃) fragments have little
internal excitation and the partner H atoms (which, by energy conservation, appear
in the early part of the TOF spectrum) show predominantly at θ = 90◦ . Expressed
in terms of the classical anisotropy parameter, β, these fragments show an angular
distribution with β ∼ −1, characteristic of dissociation following a ‘perpendicular’
photoexcitation process. At the other extreme, molecules that approach the CI with
markedly non-planar geometries have to be drawn into the deep well associated with
the CI in order to acquire the necessary planar (or near planar) geometry required
for à → X̃ surface crossing and dissociation. Inevitably, this provides the nuclei
with an out-of-plane acceleration and also tends to amplify any existing out-of-plane
nuclear motion. This angular momentum carries over into the asymptotic products
in the form of a-axis rotation in the NH2 (X̃) fragments and counterbalancing orbital
Phil. Trans. R. Soc. Lond. A (1997)
1666
M. N. R. Ashfold and others
angular momentum of the H atom about the NH2 moiety. Hence the observation
that the (slow) H atoms that partner the most highly rotationally excited NH2 (X̃)
fragments are thrown out of the original plane of the excited state molecule and,
in fact, appear with greatest probability at θ ∼ 0◦ , implying a positive β—a result
that, at first sight, might suggest that the fragments arise as a result of a ‘parallel’
photoexcitation process!
The TKER spectra shown in figures 1 and 2 also exhibit an underlying continuum,
which generally becomes more pronounced at low recoil energies, is more evident in
the case of dissociation from the 21 level than from the 00 level of the à state, and is
most evident in those dissociations which yield the asymmetric NHD fragment. As
figure 3 shows, the shape of the most dramatic of these underlying contributions, that
observed for the fragmentation NHD2 → D + NHD, is well described by a statistical
distribution based on a harmonic oscillator plus rigid rotor density of states for the
three vibrational and three rotational degrees of freedom of the NHD product, and
a factor of (Ekin )1/2 for the relative translational motion of the recoiling fragments.
Such a ‘statistical’ energy disposal has been attributed to the unimolecular decay
of highly vibrationally excited ground-state molecules formed following IC from the
initially populated à state. Before addressing further the mechanism of this à → X̃
internal conversion process in ammonia, we first consider the fragmentation of Ã
state methylamine molecules, to see if it can offer additional insights.
(b ) UV photolysis of methylamine
Figure 4 shows TKER spectra of the H + CH4 N fragments resulting from CH3 NH2
photolysis at three arbitrarily chosen, quite widely separated, near UV wavelengths:
233.3, 225.4 and 219.0 nm. The spectra recorded at the longer excitation wavelengths
show some resolved vibrational structure. Each is characterized by a rising ‘tail’ at
low TKER—indicative of some branching into fragments with high internal energy.
Studies of the relative H/D atom yields resulting from UV photolysis of each of the
isotopomers CH3 NH2 , CD3 NH2 , CH3 ND2 and CD3 ND2 , the form of the respective
H/D TOF spectra and their invariance to laser pulse energy indicate that most,
if not all, of the observed H(D) atoms arise as a result of N–H (N–D) bond fission
(Reed et al. 1996)—consistent with the earlier estimates of the relative importances of
fragmentation channels (1.3) and (1.4) (Michael et al. 1963). The maximum fragment
TKER observed in each case, indicated by the vertical arrows in figure 4, implies
D0 (H–NHCH3 ) = 34 550 ± 200 cm−1 (Reed et al. 1996).
The overall shape of these TKER spectra accords with that deduced in another
PTS study of the products arising in the 222 nm photolysis of CH3 NH2 and some of
its isotopomers (Waschewsky et al. 1995). The total CH4 N fragment yield was shown
to be isotropic (within the experimental uncertainty). This PTS study involved mass
spectrometric product detection and so was also able to investigate gross features
of the energy disposal associated with fragmentation channels (1.1) and (1.2). The
TKER spectra of the H2 elimination products from channel (1.1) and of the products
arising from the C–N bond fission (1.2) were both found to peak at TKER 0,
consistent with the presence of activation barriers in the reaction coordinates for the
corresponding back reactions (Waschewsky et al. 1995).
The present work is only sensitive to dissociations yielding H/D atoms (i.e. channels (1.3) and (1.4)) and from hereon we focus attention on these two possible primary
processes. Ab initio calculations (Dunn et al. 1996) show that à state methylamine
molecules correlate with the excited products H + CH2 NH2 (Ã)—a process that is
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
1667
signal / arb. units
(a)
(b)
(c)
TKER / cm–1
Figure 4. TKER spectra (derived assuming that the observed H atoms recoil from a partner
fragment with mass 30 amu) obtained from CH3 NH2 photolysis at (a) 233.3 nm (42854 cm−1 ),
(b) 225.4 nm (44357 cm−1 ) and (c) 219.0 nm (45648 cm−1 ) with, in each case, ε aligned
at 90◦ to the TOF axis. The arrows indicate the maximum possible TKER assuming
D0 (H–NHCH3 ) = 34 550 cm−1 . The dashed curves in (a) and (b) show the predicted form
of the respective TKER spectra if a ‘statistical’ energy partitioning involving just NH bond
fission is assumed.
endoergic for all excitation wavelengths used in this study. However, the N–H dissociation coordinate is seen to be very reminiscent of that in ammonia involving, first,
a small potential barrier at short RN−H bond extensions and, at larger RN−H , a CI
with the ground-state surface at geometries for which the NH2 moiety and one of the
C–H bonds are coplanar. The calculations indicate a barrier height of ca. 3000 cm−1
and that electronically excited CH3 NH(Ã) fragments are only energetically allowed
Phil. Trans. R. Soc. Lond. A (1997)
1668
M. N. R. Ashfold and others
for λ < 210 nm. If reliable, these values imply that even the energy provided by our
longest photolysis wavelength (233.3 nm) is just sufficient that, if appropriately redistributed, Ã state CH3 NH2 molecules are able to evolve over, rather than through,
the potential barrier in the exit channel. They also indicate that, for the present experiments, ground-state CH3 NH fragments are the only thermodynamically allowed
products that can result from the N–H bond fission process. We surmise that, as in
ammonia, the effect of the Ã/X̃ CI will be to channel any out-of-plane motion in the
excited parent molecule into fragment internal (particularly rotational) excitation.
Such an expectation is consistent with all of the experimental observations, namely:
(i) the TKER spectra obtained at the longest excitation wavelength (figure 4a) indicates some formation of CH3 NH(X̃) fragments with no internal excitation, but
the majority are formed with Eint 0; (ii) the separations (ca. 900 cm−1 ) of the
‘steps’ apparent in the TKER spectra recorded at the longer excitation wavelengths
are consistent with that expected for the in-plane HNC bending vibration in the
CH3 NH fragment (Dyke et al. 1989)—a mode which, on Franck–Condon grounds
and by analogy with NH3 photolysis, we would expected to be active during the
dissociation process; and (iii) the proportion of internally ‘cold’ fragments declines
as the excitation energy increases (see figure 4).
The TKER spectra shown in figure 4 and elsewhere (Reed et al. 1996) all appear
bimodal, with a second maximum peaking at low kinetic energies. Our studies involving the various isotopomers of methylamine indicate that these slow H/D atoms
also arise via dissociation channel (1.4). Such a result contradicts the suggestion of
Waschewsky et al. (1995), who favoured channel (1.3) as the source of these slow
H atoms, though, like these authors, we assume that these H atoms arise from unimolecular decay of highly vibrationally excited ground-state molecules (henceforth
represented as CH3 NH#
2 ) formed by IC from the initially populated à state. The
dashed curves in figures 4a, b illustrate the form of the TKER spectra that would be
expected for the H+CH3 NH products arising in the 233.3 and 225.4 nm photolysis of
CH3 NH2 if we assume (i) complete vibrational energy randomization in the CH3 NH#
2
molecules and (ii) that the probability of forming fragment pairs with any particular
total kinetic energy, Ekin , is thus proportional to the total density of product states,
ρ, at that energy. For simplicity, this ρ(E) function is modelled as the product of
just the CH3 NH fragment vibrational state density N (Ev ), at each energy Ev above
the dissociation threshold, and a term (Ekin )1/2 (where Ekin = E − Ev ) representing the three-dimensional translational density of states of the recoiling fragments
(Reed et al. 1996). N (Ev ) is calculated using the Whitten–Rabinovitch approximation (Robinson et al. 1972) together with the harmonic normal mode frequencies
listed in table 1. Rotational excitation is ignored in these calculations. The model
calculations support the premise that the observed TKER spectra should be viewed
in terms of a ‘direct’ or ‘dynamical’ component (associated with N–H bond extension
on the à state surface and coupling via the CI to the H+CH3 NH(X̃) asymptote) and
a ‘statistical’ component resulting from unimolecular decay of CH3 NH#
2 molecules.
Further, they suggest that the two routes make comparable contributions to the
total H atom yield following excitation at these wavelengths, though we concede
that a more reliable estimate of this branching requires TKER spectra displaying
a better signal to noise ratio and taken at more than one photolysis laser polarization to confirm the small angular anisotropy reported following photolysis at 222 nm
(Waschewsky et al. 1995), and that β is insensitive to the product kinetic energy.
The finding that channel (1.4) provides the major contribution to the yield of
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
1669
Table 1. Normal mode vibrational frequencies of CH3 NH2 and the CH3 NH fragment used in the
RRKM calculations described in this work
normal mode frequencies (cm−1 )
CH3 NH2 a
CH3 NH b
a
b
3361, 2961, 2820, 1623, 1473, 1430, 1130, 1044, 780,
3427, 2985, 1485, 1455, 1195, 268
3293, 2924, 2826, 1450, 1402, 1292, 1018, 964,
964, 2867, 1460, 966, 266
Motte–Tollet et al. (1992).
All reduced by 10%, cf. the ab initio values reported by Dyke et al. (1989).
slow H atoms arising via the ‘statistical’ fragmentation pathway is significant. If
we assume complete energy randomization in the CH3 NH#
2 molecules formed as a
result of IC from the initially excited à state, then RRKM theory should provide a
reasonable estimate of the energy dependent rate constant, k(E), for unimolecular
decay via the bond fissions (1.2)–(1.4). This can be estimated, for each dissociation
path and for each excess energy (E − D0 ), via the standard expression (Robinson et
al. 1972)
k(E) =
E
X
P (E − D0 )
E=D0
1
,
hN (E)
(3.2)
where the numerator represents the number of vibrational states in the respective
transition states with energy E > D0 and N (E) is the vibrational state density of
the parent CH3 NH2 molecule at internal energy E. The respective sum of states
terms for the transition states leading to C–N, C–H and N–H bond fission can be
calculated using the Whitten–Rabinovitch approximation (Robinson et al. 1972), i.e.
Ev
X
Ev =0
P (Ev ) =
(Ev + aEz )s
Q
,
s! hνi
(3.3)
where Ev is the fragment vibrational energy measured from the zero-point energy
Ez (both calculated using the normal mode frequencies listed in table 1), a is
parametrized as in Reed et al. (1996), the respective bond strengths are as listed
in the Introduction and s (= 14) is the total number of vibrational modes bar the
dissociation coordinate. We assume that twelve of the normal mode frequencies in
each transition state retain their ground-state values (see table 1). For the transition state associated with N–H fission, the NH2 asymmetric stretch (ν = 3427 cm−1 )
becomes the reaction coordinate, whilst the wavenumbers of the NH2 deformation
and the NH2 twist (ν = 1623 and 1455 cm−1 , respectively) are both arbitrarily reduced to 200 cm−1 . The transition states associated with C–H and C–N bond fission
are treated similarly. For C–H bond fission, the C–H asymmetric stretch vibration
(ν = 2961 cm−1 ) becomes the reaction coordinate and the wavenumbers of the CH3
deformations (1485 and 1473 cm−1 ) are reduced to 200 cm−1 ; for C–N bond fission,
the C–N stretch (ν = 1044 cm−1 ) disappears, the NH2 twist (ν = 1455 cm−1 ) is
reduced to 200 cm−1 and the torsional frequency is lowered to 30 cm−1 . Figure 5
shows a plot of the unimolecular decay rate constants, k(E), so derived, and of the
energy dependence of the ratio kN−H /ktotal . Clearly, in the limit of complete energy
Phil. Trans. R. Soc. Lond. A (1997)
1670
M. N. R. Ashfold and others
Figure 5. Calculated unimolecular decay rate constants, k(E), for the C–N (1.2), C–H (1.3) and
N–H (1.4) bond fission channels assuming complete energy randomization within the CH3 NH#
2
molecules before dissociation. The inset shows that under such circumstances N–H bond fission
would make, at most, a minor contribution to the total dissociation yield at all UV excitation
energies.
randomization within the CH3 NH#
2 molecules before dissociation, and irrespective of
the precise frequencies we assume for the various transition state modes of vibration,
RRKM theory predicts only a minor role for the N–H bond fission process.
Such a conclusion contradicts the experimental observations. The H atom kinetic
energy distributions observed from photodissociation of CH3 NH2 and CD3 NH2 are
very similar, whilst the D atom yield from the latter is small (Reed et al. 1996).
Similar comments apply for the D atom kinetic energy distributions observed from
CH3 ND2 and CD3 ND2 photolysis. The earlier quantum yield estimates (Michael et
al. 1963) ascribed ca. 75% of the total dissociation to N–H bond fission. None of
these results accord with the predictions arising from the RRKM modelling outlined
above. Indeed, in the limit of complete energy randomization within the CH3 NH#
2
molecules before dissociation, we should expect C–N bond fission to make the greatest contribution to the dissociation products arising from these ‘hot’ ground-state
molecules (see figure 5)—yet the quantum yield of channel (1.2) is less than 0.05
(Michael et al. 1963). Further, Waschewsky et al. (1995) show that the TKER spectrum of the small yield of CH3 + ND2 fragments resulting from 222.0 nm photolysis
of CH3 ND2 peaks away from zero, suggesting that they arise as a result of direct
dissociation on the à state surface, and not from unimolecular decay of hot CH3 ND#
2
molecules. Thus, we conclude that the dissociation of these highly vibrationally excited CH3 NH#
2 molecules leads to a democratic sampling of that portion of the phase
space associated with the H+CH3 NH dissociation pathway, but not of the total available phase space. We take this to indicate that IC to the ground-state surface occurs
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
1671
not in the vertical Franck–Condon region, but only after the occurrence of some (irreversible) N–H bond extension—most probably in the region of the conical section
where we can expect the Ã/X̃ coupling matrix elements to be largest. We may also
speculate that the large mass ratio between C or N compared with H or D inhibits
facile randomization between the —NH2 and H3 CN— vibrational modes.
(c ) Ammonia photolysis: the ‘statistical’ channel
Our earlier consideration of this channel (Mordaunt et al. 1996b) started with the
premise that IC from the 00 and 21 levels of à state ammonia molecules occurred
in the inner well and thus constituted another population loss process (additional
to tunnelling) which could contribute to the measured (lifetime limited) parent absorption linewidths. However, given the foregoing interpretation of the methylamine
TKER spectra, involving IC in the outer well associated with the Ã/X̃ CI, it is worth
considering whether a similar model could explain the ‘statistical’ contribution to the
H/D atom TOF spectra resulting from photolysis of the various isotopomers of ammonia. The measured total population loss rates (Henck et al. 1995) from both the
00 and 21 levels increase in the order ND3 < ND2 H < NDH2 < NH3 , consistent with
H(D) atom loss via a tunnelling mechanism. So, too, is the observation that the H/D
product ratio from dissociation of the mixed isotopomers is always much greater
than unity (Mordaunt et al. 1996b). For each isotopomer, population in the 00 level
decays faster than from the corresponding 21 level. This reflects the fact that the
exit channel barrier is smallest, and tunnelling consequently most facile, at planar
geometries: the wavefunction for the 21 level has a node at planarity, so molecules in
this level have to penetrate a larger effective barrier in order to dissociate (Ashfold
et al. 1985).
IC rates will depend on the background density of states and the strength of the
matrix elements coupling the à and X̃ states. The vibrational state densities of
NH3 (X̃) and ND3 (Ã) at the energy of the state origin are, respectively, ca. 40 cm−1
and 200 cm−1 (in the harmonic approximation). However, nuclear permutation symmetry must be conserved during the IC process and, as a result, for any given à state
level, the effective background state density for both NH3 and ND3 is reduced six
fold (Mordaunt et al. 1996b). NH2 D and NHD2 have lower symmetry; in this case,
permutation symmetry requirements only introduce a two-fold reduction in the total
background vibrational state density (ca. 100 cm−1 ). Clearly, density of states arguments alone cannot account for the isotopic dependence of the à state population
loss rates but the larger ‘effective’ density of receptor states in the case of NH2 D and
NHD2 is consistent with the larger ‘statistical’ contribution to the TKER spectra
obtained when photolysing these molecules (see figures 1 and 2). Progressing this
argument further is difficult without some knowledge of the way the à → X̃ coupling matrix elements vary with isotopic substitution, but the balance of evidence
must favour tunnelling through the barrier in the N–H(D) exit channel as the dominant population loss mechanism from the inner well of the à state potential energy
surface.
Thus, by analogy with methylamine, we now consider a model where the process we
have termed IC, leading to a ‘statistical’ product energy disposal, occurs in the region
of the Ã/X̃ conical intersection. Specifically, we propose that outgoing trajectories
which pass through the CI on the first encounter evolve without impediment towards
the H + NH2 (X̃) asymptote, i.e. after tunnelling through the exit channel barrier,
these dissociations are essentially ‘direct’. The topology of the parent à and X̃ state
Phil. Trans. R. Soc. Lond. A (1997)
1672
M. N. R. Ashfold and others
potential energy surfaces along the N–H dissociation coordinate dictates the form of
the excess energy disposal associated with this ‘dynamical’ fragmentation pathway
(substantial a-axis rotation of the fragment) and the product state dependent µ–
v correlations (Dixon 1996; Mordaunt et al. 1996a, b). The magnitudes of the β
parameters determined for the extremes of maximum and minimum product rotation
are also consistent with direct passage through the CI since, as discussed below,
molecules which survive in this outer well for several vibrational periods are unlikely
to yield particularly anisotropic product recoil distributions.
Now consider that fraction of the outgoing trajectories that fail to pass through
the CI on the first pass. The systems studied in this work all have insufficient energy to reach the H + NH2 (Ã) asymptote and thus must be reflected back towards
shorter RN−H bond lengths. Clearly, if the associated trajectories pass through the
CI during this compressive stage, momentum conservation dictates that they will
be drawn into the deep X̃ state well. Here they can explore much of the available
phase space and substantial intramolecular vibrational redistribution (IVR) is to be
expected. This, we propose, is the mechanism, hitherto termed IC, which leads to the
underlying ‘statistical’ component in the experimentally observed TKER spectra. As
discussed previously, one key requirement for à → X̃ transfer is planarity. Thus it
is pertinent to note that, of all the fragmentations studied, it is the dissociation of
ND3 molecules following excitation to their 00 level that yields the highest ratio of
‘dynamical’ to ‘statistical’ fragment energy disposal. It is this isotopomer that, in its
zero point level, has the smallest out-of-plane vibrational amplitude and for which the
barrier tunnelling process will provide the greatest discrimination in favour of planar
dissociations and thus passage through the CI on the first (extensive) encounter.
At the opposite extreme, consider the TKER spectrum for the D + NHD products
arising from the dissociation of NHD2 molecules following excitation to their 21
level (figure 3). The present model assumes that all of these fragmentations start
by tunnelling out of the inner well; thus the bulk of the NHD2 molecules will start
to dissociate by extending RN−H rather than RN−D . Of the former, those that pass
directly through the CI will yield H+ND2 products, with a dynamical energy disposal
(figure 2b). However, any reflected trajectories which traverse the CI whilst RN−H
is contracting will, again, have the chance to explore large portions of the X̃ state
surface and redistribute the vibrational energy initially concentrated in the N–H
bond amongst all of the available degrees of freedom. Density of state considerations
suggest that the ultimate decomposition of these ‘hot’ ground-state molecules will
yield either an H or D atom, with relative probabilities of ca. 1:2 (Mordaunt et al.
1996b). Both of these pathways should exhibit ‘statistical’ TKER spectra. Hence
the continuum underlying the structured ‘dynamical’ contribution to the H + ND2
product yield spectrum (figure 2e) and, most tellingly, the observation (figure 3) that
the TKER spectrum of the D + NHD products is almost entirely ‘statistical’; the
very small structured contribution superimposed on this continuum is attributable
to the small fraction of parent molecules that dissociate by D atom tunnelling and
direct passage through the conical intersection in the RN−D dissociation coordinate.
4. Conclusion
The bimodality of the TKER spectra derived from TOF measurements of the
H/D atoms resulting from near UV photolysis of methylamine and various of its isotopomers is consistent with there being both ‘dynamical’ (higher kinetic energy) and
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
1673
‘statistical’ (slower) contributions to the total H+CH4 N dissociation yield. N–H bond
fission is shown to be responsible for the bulk, if not all, of the H atom yield. Both
contributions are considered to arise as a result of one H atom tunnelling through
(or passing over) an earlier barrier in the N–H dissociation coordinate and evolving into the region of the conical intersection between the à and X̃ state surfaces.
‘Dynamical’ energy disposal is associated with those molecules which pass directly
through this CI and evolve out to the ground-state (H + CH3 NH(X̃)) asymptote,
whereas the ‘statistical’ contribution is attributed to those molecules that ‘miss’ the
CI on the first encounter, but make the à → X̃ transfer on a later encounter. This
allows time for possible IVR on both the à and X̃ state surfaces before dissociation
and an apparently ‘statistical’ energy disposal.
Such an interpretation has triggered further consideration of the form of the TKER
spectra derived from TOF measurements of the H/D atom products arising in the
dissociation of various isotopomers of ammonia following excitation to the 00 and 21
levels of their respective à states (Mordaunt et al. 1996a, b). A similar model which
associates ‘dynamical’ energy disposal with those molecules that pass through the
Ã/X̃ CI during bond extension and ‘statistical’ kinetic energy release with those that
transfer during NH(D) bond compression appears to provide a qualitative explanation for the way the observed H and/or D atom yields and their associated TKER
spectra vary with excitation wavelength (00 versus 21 band excitation) and isotopic
composition.
We are grateful to the EPSRC and NERC for their support of this research, to colleagues in
Universität Bielefeld, Germany (Professor K. H. Welge and Dr L. Schnieder) and in Bristol (Dr
A. J. Orr-Ewing, Dr S. H. S. Wilson, Dr C. M. Western and Dr K. N. Rosser) for their many
varied and important contributions to this programme and to Professor K. Lehmann (Princeton)
for helpful correspondence. M.K. thanks the Japanese Society for the Promotion of Science for
the award of a Postdoctoral Research Fellowship.
References
Ashfold, M. N. R., Bennett, C. L. & Dixon, R. N. 1985 Chem. Phys. 93, 293–306.
Ashfold, M. N. R., Bennett, C. L. & Dixon, R. N. 1986 Faraday Disc. Chem. Soc. 82, 163–175.
Ashfold, M. N. R., Dixon, R. N., Irving, S. J., Koeppe, H.-M., Meier, W., Nightingale, J. R.,
Schnieder, L. & Welge, K. H. 1990 Phil. Trans. R. Soc. Lond. A 332, 375–386.
Ashfold, M. N. R., Mordaunt, D. H. & Wilson, S. H. S. 1996 In Advances in photochemistry (ed.
D. C. Neckers, D. H. Volman & G. von Bünau), vol. 21, pp. 217–295 (and references therein).
New York: Wiley.
Ball, S. M., Hancock, G. & Winterbotton F. 1995 Faraday Disc. Chem. Soc. 100, 215–227.
Biesner, J., Schnieder, L., Schmeer, J., Ahlers, G., Xie, X., Welge, K. H., Ashfold, M. N. R. &
Dixon, R. N. 1988 J. Chem. Phys. 88, 3607–3616.
Biesner, J., Schnieder, L., Ahlers, G., Xie, X., Welge, K. H., Ashfold, M. N. R. & Dixon, R. N.
1989 J. Chem. Phys. 91, 2901–2911.
Dixon, R. N. 1996 Mol. Phys. 88, 949–977.
Douglas, A. E. 1963 Faraday Disc. Chem. Soc. 35, 158–174.
Dunn, K. M. & Morokuma, K. 1996 J. Phys. Chem. 100, 123–129.
Dyke, J. M., Lee, E. P. F. & Zamanpour Niavaran, M. H. 1989 Int. J. Mass Spectrom. Ion
Processes 94, 221–235.
Henck, S. A., Mason, M. A., Yan, W.-B., Lehmann, K. K. & Coy, S. L. 1995 J. Chem. Phys.
102, 4772–4782, 4783–4792.
Hwang, S. M., Higashihara, T., Shin, K. S. & Gardiner Jr, W. C. 1990 J. Phys. Chem. 94,
2883–2889.
Phil. Trans. R. Soc. Lond. A (1997)
1674
M. N. R. Ashfold and others
Kassab, E., Cleghorn, J. T. & Evleth, E. M. 1983 J. Am. Chem. Soc. 105, 1746–1753.
Lara, L. M., Lorenz, R. D. & Rodrigo, R. 1994 Planet. Space Sci. 42, 5–14.
Mack, J. A., Mikulecky, K. & Wodtke, A. M. 1996 J. Chem. Phys. 105, 4105–4116 (and references
therein).
McCarthy, M. I., Rosmus, P., Werner, H.-J., Botschwina, P. & Vaida, V. 1987 J. Chem. Phys.
86, 6693–6700.
Michael, J. V. & Noyes, W. A. 1963 J. Am. Chem. Soc. 85, 1228–1233.
Miller, R. L., Suits, A. G., Houston, P. L., Toumi, R., Mack, J. A. & Wodtke, A. M. 1994 Science
268, 1831–1838.
Mordaunt, D. H., Lambert, I. R., Morley, G. P., Ashfold, M. N. R., Dixon, R. N., Western, C.
M., Schnieder, L. & Welge, K. H. 1993 J. Chem. Phys. 98, 2054–2065.
Mordaunt, D. H., Ashfold, M. N. R. & Dixon, R. N. 1996a J. Chem. Phys. 104, 6460–6471.
Mordaunt, D. H., Dixon, R. N. & Ashfold, M. N. R. 1996b J. Chem. Phys. 104, 6472–6481.
Morley, G. P., Lambert, I. R., Ashfold, M. N. R., Rosser, K. N. & Western, C. M. 1992 J. Chem.
Phys. 97, 3157–3165.
Morley, G. P., Lambert, I. R., Mordaunt, D. H., Wilson, S. H. S., Ashfold, M. N. R., Dixon, R.
N. & Western, C. M. 1993 J. Chem. Soc. Faraday Trans. 89, 3865–3875.
Motte-Tollet, F., Hubin-Franskin, M.-J. & Collin, J. E. 1992 J. Chem. Phys. 97, 7314–7322.
Reed, C. L., Kono, M. & Ashfold, M. N. R. 1996 J. Chem. Soc. Faraday Trans. 92, 4897–4904.
Robinson, P. J. & Holbrook, K. A. 1972 Unimolecular reactions. New York: Wiley-Interscience.
Rosmus, P., Botschwina, P., Werner, H.-J., Vaida, V., Engelking, P. C. & McCarthy, M. I. 1987
J. Chem. Phys. 86, 6677–6692.
Runau, R., Peyerimhoff, S. D. & Buenker, R. J. 1977 J. Mol. Spectrosc. 68, 253–268.
Takahashi, K., Kishigama, M., Matsumi, Y., Kawasaki, M. & Orr-Ewing, A. J. 1996 J. Chem.
Phys. 105, 5290–5293 (and references therein).
Taylor, D. P. & Bernstein, E. R. 1995 J. Chem. Phys. 103, 10453–10464.
Tsuboi, M., Hirakawa, A. Y. & Kawashima, H. 1969 J. Mol. Spectrosc. 29, 216–229.
Vaida, V., Hess, W. & Roebber, J. L. 1984 J. Phys. Chem. 88, 3397–3400.
Waschewsky, G. C. G., Kitchen, D. C., Browning, P. W. & Butler, L. J. 1995 J. Phys. Chem.
99, 2635–2645.
Discussion
M. J. J. Vrakking (Department of Chemistry, Amsterdam University, The Netherlands). Professor Ashfold presented beautiful results applying the hydrogen atom
Rydberg time-of-flight technique introduced by Professor Karl Welge to studies of
molecular photodissociation. Could he give us his opinion concerning the prospects
for this technique in experiments with atoms other than hydrogen, or even small
molecules?
M. N. R. Ashfold. Several factors contribute to the exquisite resolution and sensitivity of the H(D) atom photofragment translational spectroscopy (PTS) technique
(Ashfold et al. 1996). The relevant n = 2 ← n = 1 and high n ← n = 2 transition cross sections in atomic H(D) are particularly large, so H(D) Rydberg atoms
can be prepared with high efficiency using relatively modest laser powers, thereby
minimizing any unintentional (and undesirable) competing multi photon ionization
processes. The lightness of the H(D) fragment offers a second unique advantage,
which will persist even if it eventually proves possible to prepare suitable long-lived
high-n Rydberg states of other heavier atoms or molecules with adequate efficiency.
The H(D) atom probe laser is set at the centre of the n = 2 ← n = 1 Doppler profile
Phil. Trans. R. Soc. Lond. A (1997)
Near ultraviolet photolysis of ammonia and methylamine
1675
and we detect H(D) atoms that recoil at right angles to the parent beam direction.
The recoil velocity, vrecoil , of all but the very slowest H(D) atomic fragments is so
much greater than the parent beam velocity, vparent , that the H(D) atom TOF spectrum recorded in the laboratory frame is essentially that which would be recorded
if it was possible to perform the experiment in the centre of mass (CM) frame. No
frame transformation is necessary, therefore, and the ultimate TKER spectrum has
the full energy resolution of the original data. It is worth emphasizing that comparable photofragment kinetic energy resolution can also be obtained in suitably designed
experiments in the limit that vparent vrecoil , as in the fast ion beam studies of, for
example, Neumark and colleagues (1996). In the more usual situation involving relatively heavy fragments and vrecoil ∼ vparent , it is necessary to make measurements
at several laboratory scattering angles and iterate a (inevitably cruder) trial CM
recoil distribution until it can account for all the observed laboratory frame data.
Thus, whilst many exciting possibilities can be envisaged as Rydberg (or metastable
(Morgan et al. 1996)) tagging schemes are gradually extended to a wider range of
atomic and molecular species it is not expected that such methods will provide fragment TKER spectra with the resolution or sensitivity achieved with the H(D) PTS
technique.
R. N. Dixon. Dr Vrakking asked whether the proposed distinction between dynamic and statistical routes to the dissociation of ammonia and its isotopomers
might be confirmed by time domain measurements, and would like to know the
likely timescales. I have made a time-dependent quantum mechanical study of these
processes for NH3 (Dixon 1996) using the best available potential energy surfaces for
the à and X̃ states. These surfaces were restricted to Cs geometry, with three active
vibrational coordinates (one NH bondlength, the deformation of the opposite HNH
angle, and the out-of-plane coordinate). The remaining vibrational coordinates (two
NH bondlengths and an in-plane angle) were frozen at their equilibrium values. The
timescale of the dynamical channel should be little affected by this restriction, but the
lifetime of statistical decay via the ground state surface is probably underestimated
by these calculations.
The dissociation via the first two vibronic levels of the excited state is controlled
by tunnelling through the small barriers in each exit channel, and the rate is very
sensitive to the assumed barrier height. The calculations predict liftimes of ca. 1 ps
(which is considerably longer than an estimate of ca. 200 fs from observed linewidths),
with little apparent delay between the direct and statistical components of the NH2
population distribution. An RRKM estimate of the lifetime for NH3 ‡ → H + NH2
at the available energy of 9000 cm−1 is rather less than 1 ps. These rates will be
isotope dependent. Even so, the calculations suggest that tunnelling will be the
rate-determining step. Thus it is doubtful that there will be any clear distinction
between the rates of decay through the dynamical and statistical channels, although
the statistical decay may be delayed up to 1 ps.
I. M. Mills (Department of Chemistry, University of Reading, UK). I know that
vibrational density-of-states functions are difficult to calculate accurately for an anharmonic potential energy surface for energies approaching dissociation. Can Professor Ashfold comment on whether this was a significant problem in his work, and
describe how he did this calculation?
M. N. R. Ashfold. For different reasons this was not a significant problem in modelling either NH3 or CH3 NH2 fragmentation. The D + NHD product state density
Phil. Trans. R. Soc. Lond. A (1997)
1676
M. N. R. Ashfold and others
shown in figure 3 was based on the harmonic oscillator, rigid rotor approximation
in which each vibration was treated individually whereas the rotational and translational contributions to the total state density were treated semi-classically. The
energy range involved lies well below any dissociation limit of the NHD fragment,
thus neglect of anharmonicity is unlikely to introduce significant error in this calculation. Rather, we expect neglect of centrifugal distortion effects to be the greatest
omission in this calculation, resulting in an underestimate of the total product state
density at low TKER.
The calculated unimolecular decay rates of CH3 NH2 via C–N, C–H and N–H bond
fission (figure 5) require estimation of the total density of states associated with the
various transition states. Obviously, considerable uncertainties attach to (at least
some of) the frequencies associated with the various transition states. Even so, as
discussed in our paper, the use of the Whitten–Rabinovitch approximation and the
neglect of the rotational degrees of freedom are unlikely to invalidate the overall
conclusion that N–H bond fission should be, at best, a minor fragmentation channel
for CH3 NH2 molecules in which the internal excitation is truly randomly distributed.
Additional references
Butler, L. J. & Neumark, D. M. 1996 J. Phys. Chem. 100, 12 801–12 816 (and references therein).
Morgan, C. G., Drabbels, M. & Wodtke, A. M. 1996 J. Chem. Phys. 105, 4550–4555 (and
references therein).
Phil. Trans. R. Soc. Lond. A (1997)